Identity Function and Softmax Function

An identity function outputs the input as it is. The function that outputs what is entered without doing anything is an identity function. Therefore, when an identity function is used for the output layer, an input signal is returned as-is. Using the diagram of the neural network we've used so far, you can represent the process by an identity function as shown in Figure 3.21. The process of conversion by the identity function can be represented with one arrow, in the same way in the same way as the activation function we have seen so far:

Figure 3.21: Identity function

The softmax function, which is used for a classification problem, is expressed by the following equation:

(3.10)

exp(x) is an exponential function that indicates ex (e is Napier's constant, 2.7182…). Assuming...